sinistar@mrcnext.cso.uiuc.edu (Jeff O'Hare) (10/16/90)
Hi. I was wondering if anyone has heard of any dsp hearing aids. I was think- ing about it last night, and it seems like it could be a great idea. Real-time digital equalization could maybe give users more natural hearing, correcting the natural degradation of the ear. I have heard (no pun intended) that hear- ing loss of high frequences is complete loss. Is this true, and if so, what is the approximate rolloff rate? With the DSP, such possibilities as harmon- ization could be used to lower sounds an octave so higher frequencies could still be heard, just an octave lower. Well, has anyone else thought about this, or seen any made yet? How about some dsp chips small enough to fit in a hearing aid? The 96002 is pretty large I think... Jeff
bunnell@udel.edu (H Timothy Bunnell) (10/17/90)
In article <1990Oct16.163935.1954@ux1.cso.uiuc.edu> sinistar@mrcnext.cso.uiuc.edu (Jeff O'Hare) writes: >Hi. I was wondering if anyone has heard of any dsp hearing aids. I was think- >ing about it last night, and it seems like it could be a great idea. Real-time >digital equalization could maybe give users more natural hearing, correcting >the natural degradation of the ear. I have heard (no pun intended) that hear- >ing loss of high frequences is complete loss. Is this true, and if so, what >is the approximate rolloff rate? With the DSP, such possibilities as harmon- >ization could be used to lower sounds an octave so higher frequencies could >still be heard, just an octave lower. Well, has anyone else thought about >this, or seen any made yet? How about some dsp chips small enough to fit in >a hearing aid? The 96002 is pretty large I think... > Jeff There has been an enormous amount of interest in this area and a large amount of research. There is/was at least one device already on the market (but it has not been well-received by clinicians as far as I've heard). There is research on DSP hearing aids in progress at CUNY, at the Central Institute for the Deaf, at the Research Laboratory of Electronics at MIT, and many other places. Lots of different kinds of signal processing have been tried including amplitude compression (hearing impaired people often have a very restricted dynamic range below which they cannot hear anything and above which things are painfully loud), spectral peak enhancement (amplifying only the most important parts of the instantaneous speech spectrum via adaptive filtering), frequency compression as Jeff suggested above to fit more of the speech signal into the reduced auditory bandwidth, frequency dependent amplification to compensate for the attenuation characteristics of the hearing loss, and other schemes. To date, about the only one of these that always seems to provide clear improvements in speech reception for hearing impaired people is frequency dependent amplification. Some amplitude compression is probably desirable (some of the time). The more complex schemes have tended to give sort of spotty results--sometimes promising, other times not. What all this means is that so far the one clear advantage of DSP hearing aids is their adaptability: the same piece of hardware can be programmed to the needs of many different users quite easily (given the necessary programming hardware). By contrast, analog hearing aids that shape the speech spectrum to compensate for the loss of the wearer must be prescribed and built (or at least tweeked) on a one-at-a-time basis. While this is potentially an important advantage, the economics of it have not worked out favorably yet. Analog aids are and will probably remain much cheaper for some time. Many people working in this area (including me) think that effective speech enhancement schemes will probably be based on signal processing that imitates what people do with their speech when they try to speak clearly (as opposed to their ordinary casual speech). We just haven't been clever enough to figure out how to do that automatically yet. Any new ideas out there? -- Tim Bunnell <bunnell@udel.edu>
raoul@eplunix.UUCP (Nico Garcia) (10/17/90)
In article <33682@nigel.ee.udel.edu>, bunnell@udel.edu (H Timothy Bunnell) writes: > In article <1990Oct16.163935.1954@ux1.cso.uiuc.edu> sinistar@mrcnext.cso.uiuc.edu (Jeff O'Hare) writes: > >Hi. I was wondering if anyone has heard of any dsp hearing aids. I was think- Hmmm. Why build a switched capacitor filter, or fancy processing for gain stage controls, etc., when straight forward analog ciruits will do the job efficiently and compactly? It's only audio range signals anyway, but it needs to be handled in real-time. Why ask for design sophistication when you don't need it? I also had a fascinating conversation with an audiologist here about different compression schemes. I won't go into details since I'm an EE, not an audiologist, but one scheme that works surprisingly well is clipping: map a 60 dB range speech signal into a 30 dB comfortable range for the subject by clipping the upper 30 dB right off. Some guy called Lickliter documented this technique in the 40's, and found that as long as the frequencies, relative intensity of them, and timing information was preserved, speech was still comprehensible. What it does to music, you don't even want to think about! I believe that more effective processing for speech enhancement will be based on what humans actually do, rather than what our mathematical analyses insist is most efficient or powerful. I'm not certain, however, that we understand enough of how people *hear* and process sound to approach speech enhancement from the speech production side. Note: this is my personal opinion as an educated EE, not as an audiological expert of some sort. -- Nico Garcia Designs by Geniuses for use by Idiots eplunix!cirl!raoul@eddie.mit.edu
zarko@apple.com (Zarko Draganic) (10/19/90)
There was a neat project going on when I was in college last year (I didn't get the chance to work on it). The basic idea was to apply pattern recognition techniques to the input audio signal, and recognize the voiced plosive phonemes (which are very troublesome in certain types of hearing imparments involving large high frequency attenuations). The recognized phonemes would then be substituted with "in-band" phonemes, synthesized or taken from a foreign language. Gradually the user would map the new, in-band phonemes to the lost voiced plosives. Don't know how far it's come or if anyone other than the U. of Waterloo is working on it; know of any similar efforts?
ajg@seas.gwu.edu (Alan Goldschen) (10/21/90)
In article <10810@goofy.Apple.COM> zarko@apple.com (Zarko Draganic) writes: >There was a neat project going on when I was in college last year (I >didn't get the chance to work on it). The basic idea was to apply pattern >recognition techniques to the input audio signal, and recognize the voiced >plosive phonemes (which are very troublesome in certain types of hearing >imparments involving large high frequency attenuations). The recognized >phonemes would then be substituted with "in-band" phonemes, synthesized or >taken from a foreign language. Gradually the user would map the new, >in-band phonemes to the lost voiced plosives. Don't know how far it's >come or if anyone other than the U. of Waterloo is working on it; know of >any similar efforts? I have heard of similar approach. In the 70's (or 60's), one hearing aid company designed an aid which 'mapped' frequencies that a user could not hear into ones they could hear. I cannot remember the name of the aid. I was told that the hearing aid was not comerically successful because of problems associated with training the users to understand the newer frequencies. Also, distingushing the frequencies apart was difficult. Does anybody know more about this? I would like to find out since the reasons for the aids lack of commercial success would be interesting. Thank you, ------------------------------------------------------------- Alan Goldschen e-mail: ajg@seas.gwu.edu Department of EE and CS George Washington University 801 22nd Street N.W. Washington, D.C. 20052
jbuck@galileo.berkeley.edu (Joe Buck) (10/21/90)
In article <952@eplunix.UUCP> raoul@eplunix.UUCP (Nico Garcia) writes: >I also had a fascinating conversation with an audiologist here about >different compression schemes. I won't go into details since I'm an EE, not >an audiologist, but one scheme that works surprisingly well is clipping: >map a 60 dB range speech signal into a 30 dB comfortable range for the >subject by clipping the upper 30 dB right off. Actually, one-bit speech is intelligible: take your digital speech, map all positive numbers into the same value, and all negative numbers into the same value (with opposite sign). Sounds like hell but you can understand it. >I believe that more effective processing for speech enhancement will >be based on what humans actually do, rather than what our mathematical >analyses insist is most efficient or powerful. I'm not certain, however, >that we understand enough of how people *hear* and process sound to >approach speech enhancement from the speech production side. People who do real speech processing fight this battle from both ends, and while there are many mysteries it's not as black an art as you suggest that it is. For example, in speech compression your ears really don't care if your method has the minimum mean-square error; however, if you apply a certain type of frequency weighting before you attempt to minimize the error it works quite well. To produce low-bit-rate speech of good quality you have to know a lot of DSP and ALSO know a good deal about the features of the human auditory system. -- -- Joe Buck jbuck@galileo.berkeley.edu {uunet,ucbvax}!galileo.berkeley.edu!jbuck
ajr@eng.cam.ac.uk (Tony Robinson) (10/22/90)
In article <10810@goofy.Apple.COM> zarko@apple.com (Zarko Draganic) writes: >There was a neat project going on when I was in college last year (I >didn't get the chance to work on it). The basic idea was to apply pattern >recognition techniques to the input audio signal, and recognize the voiced >plosive phonemes (which are very troublesome in certain types of hearing >imparments involving large high frequency attenuations). The recognized >phonemes would then be substituted with "in-band" phonemes, synthesized or >taken from a foreign language. Gradually the user would map the new, >in-band phonemes to the lost voiced plosives. Don't know how far it's >come or if anyone other than the U. of Waterloo is working on it; know of >any similar efforts? Well, I will be working on something similar. The basic idea is that telephone quality speech is bandlimited, 300Hz to 3.3KHz, and as a result is quite intelligible but not very natural. It is intelligible because the power spectra above 3.3KHz is mostly redundant, so it should be possible to regenerate a plausible high frequency spectra and patch the speech. This is most likely to work for unvoiced speech as then you don't have to worry about the phases of the added high frequencies, but it is unvoiced speech that suffers most from the bandlimiting anyway, so that should be okay. Maybe all this has been done before :-( Tony Robinson.
lseltzer@phoenix.Princeton.EDU (Linda Ann Seltzer) (10/25/90)
I'm not familiar with wht is available for people who have some hearing, but my understanding is that there is some DSP work going on in relation to cochlear implants. In cochlear implants, the nerve is stimulated directly, bypassing the non-functioning cochlea. For persons who do not have nerve damage or malfunction, the signals can be perceived and understood, even though they do not sound like real speech. Patients can be trained to recognize speech in these signals. DSP work is the goal of improving the signals which are transmitted to the patient, so that they produce perceived sounds which correspond more closely to speech and which are easier to interpret.
elliott@optilink.UUCP (Paul Elliott x225) (10/26/90)
For those who haven't already seen it, the latest (November? the cover
story is about electron traps) issue of Scientific American has a brief
article (near the back, I forget which section) on cochlear implants.
It mentions their history and features some newer DSP multi-channel efforts.
Interesting stuff!
Allow me to encourage the continuation of this discussion. I used
to design hearing aids (not DSP though), and find the subject fascinating.
One huge problem with DSP techniques is the pragmatic matter of operating
power requirements (high power => big, heavy, short battery life). People
with profound hearing losses will usually (but not always) accept any help
available, no matter how bulky. These cases are also where DSP methods
may be appropriate.
Has anyone managed to develop a competetive (or at least promising) DSP
design that could be implemented in an on-the-ear or in-the-ear aid? It
seems to me that the lessons learned with DSP in the lab could be best
implemented in a commercial (and cosmetically acceptable) aid using analog
techniques.
One research project I was working on (about 10 years ago) was the
implementation of multi-frequency-channel amplification and dynamic range
compression in a hearing aid using DSP methods. Not surprisingly, the
technology wasn't really ready for us to turn it into a commercially
acceptable product. How much progress has been made in the intervening
years?
Comments?
--
Paul M. Elliott Optilink Corporation (707) 795-9444
{uunet, pyramid, tekbspa}!optilink!elliott
"an archetypal entity..., superimposed on our culture by a cosmic template."raoul@eplunix.UUCP (Nico Garcia) (10/26/90)
In article <3559@idunno.Princeton.EDU>, lseltzer@phoenix.Princeton.EDU (Linda Ann Seltzer) writes: > I'm not familiar with wht is available for people who have some > hearing, but my understanding is that there is some DSP work going > on in relation to cochlear implants. In cochlear implants, the > nerve is stimulated directly, bypassing the non-functioning > cochlea. For persons who do not have nerve damage or malfunction, > the signals can be perceived and understood, even though they do > not sound like real speech. Patients can be trained to recognize > speech in these signals. DSP work is the goal of improving the > signals which are transmitted to the patient, so that they produce > perceived sounds which correspond more closely to speech and which > are easier to interpret. A good description of the process, which is what we do research on in this lab. The electrodes are implanted *in* the cochlea and stimulate the auditory nerve at different points. Since for some deaf people there is damage to the hair cells, which translate motion of fluids in the cochlea to nervous impulses, these devices essentially replace the hair cells and middle ear with electrodes and a microphone. There are two multi-channel approaches being used: the Ineraid, which has 8 electrodes and uses the output from 4 bandpass filters to stimulate a pre-selected 4 electrodes, and the Nucleus, which has 22 electrodes and stimulates different electrodes depending on the primary frequencies in the sound. The Nucleus may indeed have some digital processing, but I have heard that the builders are in the process of re-examining their stimulation scheme. If I am not mistaken, they would like to pre-select some of the channels to always carry signal, somewhat like the Ineraid. The Ineraid has *no* digital processing. (Why do DSP for a low-power compact gain control, filters, and V/I conversion? It would be like using a tennis racket to play ping-pong.) We are examining a high-speed switching scheme for the Ineraid: stimulating the electrodes sequentially with impulses from the output of their respective filters, but I would hardly call that DSP. It's still analog signals, just multiplexed in the time domain. (We expect it to reduce cross-talk.) There was a fairly good article on this stuff in a recent issue of Scientific American. If you're curious about it, you can look it up there. The nice thing about the scheme is that it takes advantage of the way the auditory nerve is already set up to take spatial locations as a spectral analysis and signal amplitude and timing information as direct information. I've had a number of arguments on why a similar process is not possible for synthetic sight. The eye has too much data processing, in my opinion, occurring before the optic nerve to pull the same sort of thing. Also, the cochlea where our electrodes is a very stable bony structure: the eye is *not*, and a fine electrode array all over the back of someone's eye does not seem reasonable. Disclaimer: I am the electrical engineer for the lab, not a clinician or audiologist. Any questions on the use or appropriateness of these devices for someone to use should be addressed to a medical expert. -- Nico Garcia Designs by Geniuses for use by Idiots eplunix!cirl!raoul@eddie.mit.edu